J . Fish Uiol. (1984) 25, 183-195

Evidence for crypsis in coho salmon, Oncorhynchus kisutch (Walbaum), parr: substrate colour preference and

achromatic reflectance W. A. DONNELLY AND L. M. DILL

Department I$ Biological Sciences, Simon Fruser University, Burnuby, British Columbia, Canada, V5A IS6

(Received 29 September 1983, Accepted 2 Nawnber 1983)

The crypsis hypothesis of salmonid parr coloration and behaviour is evaluated in light ot the criteria for protective resemblance. A review of the literature indicates that salmonid parr coloration and behaviour correspond to a cryptic interpretation. Experiments on coho salmon, Oncorhynchus kisutch, parr substrate colour preference indicate that the behavioural correlate of appropriate background choice is satisfied. Absorption spectropho- tometry of diapositives of fish and experimental substrates suggests that background match- ing is achieved through achromatic reflectance and absorption of wavelengths by the silvery sides and parr marks, respectively.

1. INTRODUCTION

The action of natural selection has been convincingly demonstrated by investi- gations of the protective resemblance of organisms to their environments, a phe- nomenon known as crypsis. Kettlewell’s (1955) study of industrial melanism in Biston betularia is probably the best known example but a wealth of literature illustrating the role of crypsis has been generated in the past hundred years. While many investigations only presume the relationship of form to function, recent studies (e.g. the role of bioluminescence in mesopelagic countershading, Young et al., 1979) have eliminated much doubt as to their validity by fulfilling Suffert’s (1 932; from Robinson, 1969) criteria of protective resemblance:

( I ) Evidence of convergence of similar characters derived from distinct mor- phological origins in different species or species groups.

(2) Evidence that the characters are correlated with the appropriate behavioural patterns.

( 3 ) Evidence that the characters perform the hypothesized task. This investigation evaluates salmonid parr coloration in light of these criteria. The adaptive significance of ‘ stream coloration ’ (Nikolsky, 1963) in temperate

freshwater fishes has received little attention, but due to its potential implications for fisheries management and enhancement deserve study. Stream coloration of fishes is characterized by dorsal countershading and the presence of discrete bars along the silvery lateral sides. In salmonids, these aggregations of chromatophores are called parr marks and occur on all juvenile Oncorhynchus (except pink sal- mon, 0. gorhuscha), and all juvenile Prosopium, Salmo, Salvelinus and Thymal- lus spp. (Scott & Crossman, 1973). Similar markings occur in many cyprinid species, for example, the common minnow, Phoxinus phoxinus (Nikolsky, 196 l) ,

183 0022-11 12/84/080183+ 13 $03.00/0 0 1984 The Fisheries Society of the British Isles

I84 W . A . DONNELLY A N D L. M . DILL

and some Asiatic barbs (Kortmulder, 1972). Many species of North American freshwater fishes (Gasterosteus, Rhinichthyes and Cottus) possess similar bars, blotches or saddles at some stage of their life histories (Scott & Crossman, 1973). This similarity in coloration comes close to satisfying Suffert’s convergence criter- ion but falls short in that all similar markings of this type are derived from the same neural crest material (Parker, 1948). We interpret the independent evol- ution of similar markings in these taxonomically diverse groups, and their predic- table absence under certain ecological conditions (see Discussion), as sufficient evidence for convergence. We suggest that parr marks confer considerable selec- tive advantage and are not merely afunctional vestiges of ancestral coloration as implied by Norman (1 93 1).

Although early authors such as Calderwood (1 907) described salmonid parr coloration as protective, only Hoar (1 958) has provided experimental evidence for this hypothesis. Of four species of Oncorhynchus parr subjected to predation by wild crows, three parr-marked species had negligible mortality but pink salmon fry sustained high losses. Hoar attributed the high survivorship of parr-marked species to adaptive antipredator behaviour and cryptic coloration. Although salmon parr are now accepted to be cryptic (Nikolsky, 1963; Denton, 1971; Hawkes, 1974; Davies & Thompson, 1976; Hasler et af., 1978; Ruggles, 1980; Noakes, 1980) no additional research has been done and no mechanism proposed for the function of parr marks. Some researchers have suggested that parr-marked fishes are most cryptic when viewed over gravel substrates (Nikolsky, 1963; Davies & Thompson, 1976). This explanation, consistent with Endler’s (1978, 1980, 1982) work on poeciliid background matching, is the hypothesis that we have adopted. Noakes (1980) suggested that parr marks resemble shadows cast by ripples at the water surface (also McFarland & Loew, 1983). Since these shadows are cast on the substrate, Noakes’ hypothesis also implies background matching. Maeda & Hidaka (1979) reported that cherry salmon, 0. masou, use parr marks in intra-specific displays of territorial aggression, but coloration may serve both cryptic and communicative purposes as a function of viewing distance (Hailman, 1977), so a cryptic interpretation of parr mark function is not incompatible with a communication hypothesis.

Crypsis is a primary defensive tactic whose success depends upon sufficient morphological resemblance of prey to a representative sample of the habitat to reduce the likelihood of predator visual recognition. Because the fidelity of this resemblance is enhanced by the appropriate behaviour, cryptic animals possess convergent behaviours resulting from similar selection pressures. When the behavioural correlates of a generalized cryptic organism (Robinson, 1969; Edmunds, 1974) are compared to salmonid parr behaviour (Table I ) a close correspondence is obtained, satisfying Suffert’s behavio.ural criterion of protective rese mb I an ce.

Restriction of movement is the first limitation placed on the behaviour of cryptic organisms. To be cryptic, animals must be motionless with respect to their backgrounds. For fish in shallow and clear fast-flowing waters, crypsis requires swimming with the current to remain inconspicuous against the stationary sub- strate. Stream dwelling salmonids conform by exhibiting positive rheotaxis even in rapid currents (Keenleyside & Hoar, 1954). Conflicts between feeding and avoiding predator detection arise with restricted movement. Essential

COHO SALMON CRYPSIS 185

TABLE 1. Behavioural correlates of crypsis

Crypsis requirements Salmonids Reference

Restricted movement Positive rheotaxis Restricted feeding Crepuscular activity

Risk sensitivity Escape behaviour Dart-freeze escape

Substrate pressing Colour change abilities Acclimational

Physiological Prevent search image Territoriality Background choice Cover and shade

Substrate choice

Keenleyside & Hoar, 1954 Butler & Hawthorne, 1968 Dill, 1983 Hoar, 1958 Hoar, 1958 Neill, 1940 Hoar, 1955 Dill ct al., 1981 Kwain & McCrimmon, 1967 Ritter & McCrimmon, 1973a.h

activities such as feeding are usually carried out when there is a reduced risk of discovery by predators. Salmonids feed most actively at dawn and dusk (Butler & Hawthorne, 1968) but some sampling of drift occurs during the day (Dill et al., I98 1). While darting to capture prey, coho contravene the movement restric- tion rule and risk visual detection by predators. Salmonid parr should therefore be sensitive to predation risk and restrict feeding when risk is high. Indeed, the reactive distance of coho parr to drift is inversely related to the presentation frequency of an artificial predator (Dill, 1983).

Cryptic animals have characteristic secondary defence tactics, and when dis- covered by predators, dart a short distance and freeze against an appropriate background. They may also press themselves against the background to minimize shadow contrast (Hailman, 1977). Salmonid parr have similar escape and hiding behaviour (Hoar, 1958). Cryptic organisms have well developed colour change abilities to blend with diverse backgrounds. Salmonid parr change colour in response to substrate brightness (Neill, 1940) and the intensity of overhead illumination (Hoar, 1955). Crypsis is vulnerable to search-image formation by predators (Tinbergen et af., 1959). Typical defences against this include poly- morphism and adaptive inter-individual spacing. No data are available on salmon parr colour polymorphisms, although pattern diversity in terms of parr mark number occurs within local stocks (Smith, 1969). Territoriality is widespread in stream dwelling salmonid parr (Dill et al., 198 1 ) and likely serves an antipredation function. Territorial Atlantic salmon, Salmo salar, experience less predation by brook charr, Sulvelinus fontinalis, than do non-territorial conspecifics (Symons, 1974).

Organisms using crypsis are required to display against appropriate back- grounds. Cryptic salmonid parr should reduce their risk of predator visual detection by holding position over substrates that minimize their visibility. Studies on yearling rainbow trout, Salmo guirdneri, have demonstrated cover and shade seeking behaviour (Kwain & McCrimmon, 1967) and preference for black versus white substrates (Ritter & McCrimmon, 1973a,b). Use of cover and shade and the preference of yearling parr for black substrates could reduce predator

I86 W . A. D O N N E L L Y A N D L . M. D I L L

detection by minimizing the apparent contrast of fish over gravel substrates. Since no experiments on salmonid parr substrate colour preference have been reported, we tested the prediction that coho parr should hold position over coloured substrates which minimize their contrast and colour mismatch.

Observations made during these colour preference experiments revealed that the silvery sides and bellies of coho parr reflect the gravel substrate colour. This ' achromatic reflectance ' (Hailman, 1977) is well documented for pelagic fishes (Denton, 1971) and is an untested hypothesis to explain the white ventral col- oration of amphibians and reptiles (Norris & Lowe, 1964). If salmonid parr utilize achromatic reflectance in a similar manner, the spectrum of reflected colours from their silvery sides should correspond to the reflectance of the gravel substrate in wavelength and intensity. The dermis of coho salmon contains iridophores, xan- thophores and melanophores (Hawkes, 1974) with green being the overall colour reflected by the iridophore-xanthophore interaction. Because melanins generally absorb light in the visible spectrum uniformly (Norris & Lowe, 1964), we pre- dicted that parr marks should absorb light reflected from the substrate, thereby decreasing the intensity of light reflected from the lateral sides of the fish. Parr marks were predicted to render coho more inconspicuous than achromatic reflec- tance alone.

11. MATERIALS AND METHODS SUBSTRATE COLOUR PREFERENCE

Coho parr 40-80 mm P.L. were obtained from Capilano Hatchery, North Vancouver, British Columbia and held outdoors in white plastic flow-through containers measuring 60 x 30 x 30 cm. Six colours of 5-10 mm gravel were obtained from commercial sup- pliers and fastened to 30 x 30 x 0.06 cm polyethylene sheets using silicone sealant. These gravel pads facilitated quick changes of substrate colour. The gravel colours are abbre- viated throughout the remainder of the text. The abbreviations and their corresponding colours are: BLK=black silica gravel; CRY =mix of black, grey and white aggregate; BRN = mix of dark brown, light brown and some white aggregate; GRN =light with some dark, green epoxy coated aggregate; GLD=light with some dark, gold epoxy coated aggregate; TAN=mid brown, tan and white epoxy coated aggregate. The BRN was the most ' natural ' colour, judging by salmon streams in south coastal British Columbia.

Experiments were conducted outdoors at the S.F.U. Animal Care Facility from May to July 1982. Two pads having different substrate coloration were placed in a 60 x 30 x 15 cm white plastic container and 100 coho parr introduced. After 15 min the fish were counted over the least preferred colour and the larger number over the preferred substrate obtained by subtraction. The substrate colours were replaced with new colours, 15 min allowed to elapse, and counts again taken. Each colour of pad had an equal probability of being on the left or right side of the container. Each substrate colour was compared to all colours but itself four times for a total of 60 trials. N o substrate colour was used more than once in any two successive trials and no group of fish was used in more than four consecutive trials. A Pentax spotmeter was used to measure the reflectance of each substrate under identical lighting conditions to separate the effects of brightness and colour per sr. Spotmeter readings were transformed to lux. Results were analysed using chi-square (a=0-05).

DORSAL REFLECTANCE To determine if the coho salmon chose the appropriately coloured background, i.e.

the one which minimized the spectral mismatch between their dorsums and the gravel, diapositives of the dorsums ofa group of 100 coho salmon were taken using a 35 mm Nikon F2 camera with 5 5 mm f3.5 macro lens and 400 ASA Kodak Ektachrome film. The

COHO S A L M O N C R Y P S I S 187

diapositives were mounted and analysed with a Carey 2 10 absorbance spectrophotometer. Similar techniques, using intact animals and reflectance spectrophotometry were used by Norris & Lowe (1964) and Lilywhite et al. (1977). Differences between diapositives were eliminated by using the auto-baseline option on the Carey 210 and comparing two diapositives of the substrate (thus making the substrate trace equal to zero throughout the visible spectrum). For the baseline measurement a scan rate of0.5 nm sr l and a beam width of 1 nm were used. A 4 x 10 mm window was cut in a piece of light cardboard and placed over an area of the diapositive where 90% of the substrate was covered by the dorsums of the coho. This area was scanned from 700-400 nm at I nm s-I. The difference in area between the spectral record obtained for the fish and the zero baseline provides an index of the degree of colour mismatch between the dorsums of the fish and the gravel substrate.

ACHROMATIC REFLECTANCE We photographed two coho salmon in a three-sided clear plexiglass chamber

( I 5 x I2 x 10 cm) with the open side propped against the 10 mm glass window of a 9000 I artificial stream channel. The fish were photographed 1-5 cm above three of the gravel colours used in the previous experiments (TAN, GRN and BRN) plus a fourth, highly reflective dayglo yellow substrate (YEL). A slide of exposed, blank diapositive was used in the reference beam of the spectrophotometer to eliminate emulsion characteristics. Reference slit width was controlled with a comb to balance the absorbance ofthe reference and sample diapositives at 700 nm. Selected regions of two diapositives per substrate colour were scanned through a 2 x 10mm window (Fig. 1 ) cut from a piece of light cardboard, at a rate of 2 nm s-I in the 700 to 400 nm waveband using a 1 nm beam width. Because the gravel substrates were not uniform in colour and brightness, two absorbance spectra were obtained from the gravel substrate [Fig. I(a)] on each diapositive to provide a range (n=4) indicative of substrate heterogeneity. One record per diapositive was obtained for the lateral sides of one fish just below the parr marks [Fig. I(b)], and in the parr marked region along the lateral line pig. I(c)]. The pairs of values obtained were averaged to provide the best estimate of reflectance for the region measured. The degree of trace superposition between the substrate and the region of the fish measured was interpreted as evidence of background matching in like manner to Norris & Lowe (1964).

To determine if the difference in absorbance between the two areas of the fish scanned was due to the presence of parr marks or to differences related to body morphology, one of the diapositives obtained for the GRN substrate was masked so that absorbance spectra could be obtained for two individual parr marks and two ofthe unpigmented areas between them.

While these data illustrate relative differences between the transmission of light through the gravel or lateral sides of the fish on the diapositives. the diapositives utilize the oxidation of azo dyes corresponding to the secondary colours cyan, yellow and magenta, rather than the primary colours blue. green and red (Ferguson. 1980). The use ofsecondary colours instead of the primary colours perceived by the cones of the vertebrate eye means that this technique is not directly analogous to vertebrate photopic vision. Nevertheless it is sufficiently similar to be useful in the solution of this and other problems involving colour vision, For example, when the green and red cones of the human retina are stimu- lated about equally and the blue cones receive minimal stimulation, the psychophysical sensation perceived is yellow (Lythgoe, 1975). For the diapositives, yellow is a ' primary ' colour but the spectrophotometer correctly identifies the yellow waveband and degree of absorbance regardless of how it is perceived.

111. RESULTS

SUBSTRATE COLOUR PREFERENCE Figure 2 gives the total number of fish recorded over each substrate colour as

well as the reflectance o f t h e various substrates. Each histogram sums the numbers from all tests using a particular colour in combination with all other colours.

188 W. A. D O N N E L L Y A N D L. M . D l L L

FIG. 1. Areas ofdiapositive scanned. Substrate (a), latcral sides excluding parr marks (b), lateral sides including parr marks (c).

I 200 7- --TI 3500

'oool l l o 0 l

L

z

800

7301

600'

t 5 0 0

-2000 :

- 1500

- 1000

500 ELK GRY E R N GRN GLD T A N

Substrole colour

FIG. 2. Substrate colour preference (bars) and substrate reflectance (triangles).

Gencrally, more absorptive (darker) were preferred to more reflective (lighter) substrates. However, the most absorptive BLK substrate was not significantly preferred to the more reflective GRY or BRN gravels. The GRN substrate had identical reflectance to the BRN substrate, yet coho preferred to position them- selves over BRN. In addition, the two most reflective substrates were significantly different in terms of coho preference, but in the opposite direction to that which would be predicted by reflectance alone.

Table I1 gives probability values for the pairwise substrate colour preference experiments. The pairwise results largely concur with the cumulative results. The BRN substrate was significantly preferred to all other substrates except C R Y , for which there was no significant difference. CRY was preferred to all substrates

COHO SALMON CRYPSIS 189

TABLE 11. P-values for substrate colour pair experiments. Pre- ferred colours are in left hand column

GLD GRN TAN BLK GRY BRN BRN 0.001 0.05 0.05 0.05 n/s X CRY 0.0 1 n/s 0.001 0.001 X BLK 0.001 0,001 0.001 X TAN n/s 0.05 X GRN 0.001 X GLD X

-0 .201

4 00 500 600 7

Wavelength (nm)

)O

Fic. 3. Absorbance spectra of coho dorsums over individual substrate colours (substrate colours are the zero baseline).

except BRN and GRN, for which there was no significant difference. BLK was preferred to all of the remaining substrates. TAN was preferred to GRN, yet no significant difference existed between TAN and GLD. GRN was preferred to GLD and GLD was not preferred to any other substrate tested. The rank order of preference is therefore BRN3 GRY > BLK> TAN > GRN> GLD. Except for the reversed order of TAN and GRN, and the lower position of BLK, pairwise rank order is similar to that derived from the orthogonal comparison.

DORSAL REFLECTANCE Because absorbance spectrophotometry was used, all figures are plotted in terms

of absorbance. An absorbance value of 1.0 corresponds to 10% transmission of the incident light through the diapositive, and a value of 2.0 to 1% transmission. Troughs in the figures that follow represent those wavebands transmitted through the diapositive (equivalent to those reflected by the fish or gravel surface) and are

I90 W . A. 1)ONNF:LLY A N D I.. M. DILL

Wavelength (nm)

FIG. 4. Absorbance spectra of mean values obtained for coho lateral sides (-) and parr marks (---) and ranges for substrates (shaded). (a) YEL substrate, (b) TAN, (c ) GRN, (d) BRN.

those most visible to the naked eye. Peaks represent wavebands absorbed and therefore relatively less visible to observers.

Figure 3 shows the spectral mismatch of coho parr dorsums over five of the gravel substrates used in the substrate preference experiments (BLK was elimi- nated from this comparison due to lack of transmission through the diapositive of the black substrate). The degree of spectral mismatch (as determined by area) was least for BRN and GRY indicating that the dorsums of coho are most cryptic over these substrates. The area calculated was similar for GRN and GLD (and about twice the area calculated for BRN and GRY) indicating a substantial mismatch with respect to these substrate colours. The area about the baseline for the dorsums of the fish in comparison to the TAN substrate is approximately three times the area for either the BRN or GRY substrate, indicating the greatest spectral mismatch.

ACHROMATIC REFLECTANCE Figure 4(aHd) gives for each substrate colour the range of substrate absorbance

and the mean values obtained for the lateral sides including and excluding the

COHO SALMON C R Y P S I S 191

I I I 400 500 600 7 0

Wavelength (nm l

FIG. 5 . Absorbance spectra of two individual spaces between parr marks along lateral sides (-) and two individual parr marks (---).

parr-marked regions. The superposition (indicated by similarities in the position of peaks and troughs) of the mean lateral side traces with the ranges provided by the four substrate colours is obvious for both parr-marked and unmarked regions. This superposition illustrates that the appropriate wavebands are being reflected by the lateral sides of the fish to render them reasonably cryptic over the four substrates.

The parr-marked region generally absorbs more strongly than the silvery sides. Over the more reflective substrates (YEL, TAN and GRN), this results in the parr-marked region matching the substrate less well than the unmarked region. However, over the more natural (and preferred) BRN substrate, the parr marked and unmarked regions of the lateral sides matched substrate absorbance equally well.

Some features of these figures deserve explanation. For Figs 4(a), (c) and (d) transmission through the diapositive is greater than 100%. This is apparently common for fluorescent colours (such as YEL) viewed in water (Luria & Kinney, 1970) and is attributable to scattering. For Figs 4(a) and (c) the mean absorbance of the unmarked lateral sides falls approximately midway between the ranges obtained for the substrates in the 520-700nm range, but for Fig. 4(b) there is greater absorbance by the lateral sides of the fish than by the substrate in the same waveband. The substantial amount of white in the TAN substrate is likely responsible for this degree of mismatch. A crossover of the lateral side and parr mark traces occurs at 520 nm in Figs 4(a) and (b). Since melanins generally absorb uniformly in all regions of the visible spectrum, this crossover was unexpected. It was also present in the substrate traces, suggesting that it is an artefact of the technique used. No crossover was present for either the GRN or the BRN substrate or the corresponding fish traces in Figs 4(c) and (d).

192 W . A . D O N N E L L Y A N D L . M . DI1.L

Figure 5 shows the absorbance of light reflected from a green substrate by two individual parr marks, and by two of the areas between parr marks. The parr marks absorb more strongly than the silvery areas between parr marks in all but the blue region of the spectrum indicating that the differences in absorbance in Fig. 4 are attributable to the parr marks per SL' and not to the locations where readings were taken.

IV. DISCUSSION

The results support the hypothesis that coho parr use crypsis for primary defence. The substrate preference experiments demonstrated that although coho generally position themselves over low reflectance substrates (similar to the findings of Ritter & MacCrimmon, 1973a,h), substrate coloration determines preference for substrates similar in reflectance (GRN and BRN). The dorsal and achromatic reflectance results demonstrate that coho parr generally prefer the substrate colours which reduce contrast and colour mismatch. The achromatic reflectance results further demonstrate that reflectance by the lateral sides of the fish falls within, or borders on the range of the samples of the gravels throughout the visible spectrum. These findings reinforce the behavioural correlate of appropriate background choice and are consistent with a cryptic interpretation of salmonid parr coloration.

Fish in pelagic habitats with fairly narrow high-energy photic conditions require only thin-layer reflecting lateral sides and countershading to achieve crypsis. In shallow nearly transparent waters, light conditions are much more variable. The variety of wavelengths transmitted through the medium and differential absorp- tion and reflection by the colours and shades of the gravel suggests that for crypsis against heterogeneous backgrounds, achromy may be insufficient because an unrepresentative (lighter) sample of the habitat would be reflected by the lateral sides. Parr marks may thus improve upon the resemblance of the fish to the substrate by absorbing light in a manner similar to the darker substrate particles.

Beyond some critical distance (the blending-distance, Mottram, I9 15) adjacent colour patches cannot be resolved. The blending distance in an aquatic habitat will be relatively shorter than in air due to differential scattering and absorption of longer wavelengths by water (Endler, 1978). Beyond this distance, the lateral sides and the parr markings might thus average to yield an intermediate shade closely paralleling the mean substrate coloration. This is especially evident over the BRN substrate, which was also most preferred by the fish.

The preceding argument may explain the absence of parr marks for species that do not associate with a gravel substrate (such as pink salmon which go to sea immediately following emergence) and the fading of parr marks on anadromous species when smoltification occurs. In pelagic habitats where the substrate is a great distance from the water surface, laterally positioned melanins can only detract from the resemblance of the fish to its habitat.

The parr marks of salmonid species tend to fade as the fish grow. This could be due to the fact that they move to deeper water (Wankowski, 1979), with differ- ent lighting conditions, or to the possibility that a refuge from predation is reached with the attainment of large body size.

Hoar (1955) showed that both the eye and pineal are responsible for the

COHO S A L M O N C R Y P S I S 193

melanophore response that results in lightening or darkening of the dorsums of juvenile sockeye salmon, 0. rzerka, according to the amount of downwelling light. We suggest that these photoreceptors also disperse or aggregate the melanin with- in the melanophores of the parr marks so that the appropriate amount of light is reflected from the silvery sides of the fish to match substrate albedo. Albedo depends upon the characteristics of the substrate and the medium as well as the incident downwelling light and would thus serve as a more comprehensive cue for melanin aggregation and dispersal.

The cryptic interpretation of salmonid parr coloration and behaviour may have wide-reaching practical implications for salmonid management and enhance- ment. While achromatic reflectance and parr marks should render laterally viewed parr-marked species cryptic over most substrate shades and colours, the dorsal coloration within stocks is unlikely to be as flexible. The dorsal coloration of any stock will likely depend upon the substrate characteristics, the wavelengths characterizing the habitat, water depth and clarity, the amount of shade and cover, the stochasticity within these variables, the ability of the stock to track the habitat, and the visual acuity and other characteristics of the predation complex within the habitat. We predict that predatQr selection for crypsis, and limited genetic mixing between stocks due to the propensity of salmonids to return to their natal streams to spawn, should result in the evolution and maintenance of local colour variants in salmonid parr. Variability in salmonid coloration posed problems for early investigators. Early taxonomists described local colour variants as new species before coloration was dismissed as a salmonid systematic character (Seeley, 1886). Recent authors (McPhail & Lindsey, 1970; Scott & Crossman, 1973) describe adult salmonid coloration as highly variable from watershed to watershed and stock to stock. If similar colour variation exists in local popu- lations of salmonid parr, and if the variation is genetically based, transplanting of fry to habitats where they did not evolve would be predicted to have only limited success. Field sampling to determine variability in dorsal coloration and parr mark characters of salmonid parr from different habitats is currently underway.

The assistance of Jeff May in collection of substrate colour preference data is gratefully acknowledged. Paul Saunders and Doug Wilson provided many helpful suggestions for the spectrophotometry techniques. This work has benefited by discussions with J. Burr, R. Dunbrack, L. Giguere, J. McPhail, K. Puckett and D. Ware. We thank J. Abbott and D. Donnelly for commenting on an earlier version of the manuscript. Financial support was provided by Canadian National Sportsman’s Fund Research Grants, and NSERC Canada grant A6869.

References Butler, R. L. & Hawthorne, V. M. (1968). The reactions of dominant trout to changes

Calderwood, W. L. (1 907). The Life of the Salmon. London: Edward Arnold. pp. 160. Davies, R. W. &Thompson, G. W. (1976). Movements of mountain whitefish (Prosopium

williamsoni) in the Sheep River Watershed, Alberta. J . Fish. Res. Bd Can. 38,

in overhead artificial cover. Trans. Am. Fish. SOC. 97, 3 7 4 1 .

2395-240 1.

194 W . A . IIONNELLY A N D L. M . DILL

Denton, E. J. (1971). Reflectors in fishes. Sci. A m . 224, 64-72. Dill, L. M. ( 1 983). Adaptive flexibility in the foraging behaviour of fishes. Can. J . Fish

"kput . Sci. 40, 398-408. Dill, L. M., Ydenberg, R. C. & Fraser, A. H. G. (1981). Food abundance and territory

size in juvenile coho salmon, 0ncorhytwhu.s kisutch. Can. J . Zool. 59, 1801-1 809. Edmunds, M. (1974). Defense in .4nimuls. Harlow, Essex: Longman. pp. 357. Endler, J. A . ( 1 978). A predator's view of animal colour patterns. In Evolutionary Biologj,:

11. (M. K. Hecht, W. C. Steere & B. Wallace, eds), pp. 3 19-364. New York: Plenum Press.

Endler, J. A. ( 1 980). Natural selection on colour patterns in Potcilia rclicziluta. Evolution

Endler, J . A. (1982). Convergent and divergent effects of natural selection on colour

Ferguson, B. (ed.) (1980). Kodak Color Films 3rd edn. Publication E-77. Rochester. New

Hailman, J. P. (1977). Optical Signals. Bloomington: Indiana University Press. pp. 362. Hasler, A. D., Scholz, A. T. & Horrall, R. M. (1978). Olfactory imprinting and homing

Hawkes, J. W. (1974). The structure of fish skin. I I . The chromatophore unit. Cell 7

Hoar, W. S. (1955). Phototactic and pigmentary responses of sockeye salmon smolts following injury to the pineal organ. J. Fish. Rcs. Bd Can. 12, 178-185.

Hoar, W. S. (1958). The evolution o f migratory behaviour among juvenile salmon of the genus Oncorhynchiis. J . Fish. Rev. Bd Cun. 15, 391-428.

Keenleyside, M. H. A . & Hoar, W. S. (1954). Effects of temperature on the responses of young salmon to water currents. Behaviour 7, 76-87.

Kettlewell. H. B. D. (1955). Recognition ofappropriate backgrounds by the pale and black phases of Lepidoptera. Nature. Lond. 175, 943-944.

Kortmulder, K . (1972). A comparative study in colour patterns and behaviour in seven Asiatic Barbus species (Cyprinidae, Ostariophysi. Osteichthyes). Behuviour Suppl.

Kwain, W. & McCrimmon, H. (1967). The behaviour and bottom colour selection of the rainbow trout, Sulmo gairdneri Richardson, exposed to different light intensities. Anim. Bchav. 15, 75-78.

Lilywhite, H. B., Friedman, G. & Ford, N. (1977). Colour matching and perch selection by lizards in recently burned chaparral. C'opeia 1977, 1 15-1 2 1.

Luria, S. M. & Kinney. J . A. S . (1970). Underwater vision. Science 167, 1454-1461. Lythgoe. J . N. (1975). Problems of seeing colours under water. In Vision in I.'isht>.s

(M. A. Ali, ed.), pp. 6 19-634. London, New York: Plenum Press. Maeda, N. & Hidaka, T. (1979). Ethological function of parr marks in a Japanese trout,

0ncorhynchu.s masou f. ishikarvai. Zool. M a g . (Tokyo) 88, 34-42. McFarland, W. N. & Loew, E. R. (1983). Wave produced changes in underwater light and

their relations to vision. In Predators and Prey in Fishes (D. L. G. Noakes, D. G. Lindquist, G. S. Helfman & J. A. Ward, eds), pp. 11-22, The Hague: Dr J. W. Junk,

McPhail, J. D. & Lindsey, C. C. (1970). Freshwater fishes of Northwestern Canada and Alaska. Fish. Res. Bd Can. Bull. 173, pp. 38 1.

Mottram, J. C. (191 5) . Some observations on pattern-blending with reference to oblitera- tive shading and concealment of outline. Proc. Zool. Soc. Lond. 47,679-692.

Neill, R. M. (1940). On the existence of two types of chromatic behaviour in teleostean fishes. J . E x p . Bid . 17, 74-98.

Nikolsky, G. V. (I96 I ) . Special Ichthyology. Jerusalem: Israel Program for Scientific Translations. pp. 538.

Nikolsky, G. V. (1963). The Ecology vf Fishes. London: Academic Press. pp, 352, Noakes, D. L. G. (1980). Social behaviour in young charrs. Per.7pective.s in Vertebrate

Science, Vol. I . C h a m , Salrnonid,fishes of the Genus Salvelinus (E. K. Balon, ed.), pp. 683-701. The Hague: Dr J. W. Junk.

34, 76-9 1 .

patterns in two fish faunas. Evolution 36, 178-1 88.

York: Eastman Kodak Co.

in salmon. Am. Sci. 66, 347-355.

RPS. 149, 159-172.

19, 1-33 I .

Norman, J. R. (1931). A History ofFishes. London: Ernest Benn. pp. 463.

COHO SALMON C R Y P S I S 195

Norris, K. S. & Lowe, C. H. (1964). An analysis ofbackground color-matching in amphib- ians and reptiles. Ecology 45, 565-580.

Parker, G. H. ( I 948). Animal Colour Changes and their Neurohumours. Cambridge: Cambridge University Press. pp. 377.

Ritter, J. A. & McCrimmon, H. R. (1973~) . Influence of environmental experience on responses of yearling rainbow trout (Salmo gairdneri) to a black and white substrate. J . Fish. Res. Bd Can. 30, 1740-1742.

Ritter, J. A. & McCrimmon, H. R. (19736). Effects of illumination on behaviour of wild brown (Salmo trutta) and rainbow trout (Salmo gairdneri) exposed to black and white backgrounds. J. Fish. Res. Bd Can. 30, 1875-1 880.

Robinson, M. H. (1969). Defenses against visually hunting predators. In Evolutionary Biology: 3 (T. Dobzhansky, M. K. Hecht & W. C. Steere, eds), pp. 225-259. New York: Meredith Corporation.

Ruggles, C. P. (1980). A review of the downstream migration of Atlantic salmon. Can. Tech. Rep. Fish. Aquatic Sci. 952, 1-39.

Scott, W. B. & Crossman, E. J. (1973). Freshwater fishes of Canada. Fish. Res. Bd Can. Bull. 184, pp. 966.

Seeley, H. G. ( 1 886). The Fresh- Water Fishes of Europe. London: Cassell. pp. 444. Smith, S. B. (1969). Reproductive isolation in summer and winter races of steelhead trout.

In Symposium on Salmon and Trout in Streams (T. G. Northcote, ed.), pp. 21-38. Vancouver: University of British Columbia.

Symons, P. E. K. (1974). Territorial behaviour of juvenile Atlantic salmon reduces predation by brook trout. Can. J . Zool. 52, 67-69.

Tinbergen, N., Impekoven, M. & Franck, D. (1959). An experiment on spacing-out as a defence against predaton. Behaviour 28,307-32 1.

Wankowski, J . W. J. (1979). Spatial distribution and feeding in Atlantic salmon Salmo salar L. juveniles. J . Fish Biol. 14, 239-247.

Young, R. E., Roper, C. F. E. & Waiters, J. F. ( I 979). Eyes and extraocular photoreceptors in midwater cephalopods and fishes: their roles in detecting downwelling light for counterillumination. Mar. Biol. 51, 37 1-380.